From the top down:

Self-esteem and self-evaluation

Jonathon D. Brown, Keith A. Dutton, and Kathleen E. Cook
University of Washington, Seattle, WA, USA
The affective model of self-esteem development assumes that: (a) self-esteem
forms early in life in response to relational and temperamental factors; and (b) once
formed, endows high self-esteem people with the ability to promote, protect, and
restore feelings of self-worth. In this article, we use the model to examine the
relation between self-esteem and self-evaluations, showing that one way high self-
esteem people maintain feelings of self-worth is by claiming to possess socially
valued qualities. We conclude by considering the implications of the affective
model for understanding the nature and functions of self-esteem.
Self-esteem is one of psychologys most popular constructs. It is used as a
predictor variable (some researchers study whether high self-esteem people
think, feel, and behave differently than do low self-esteem people), an out-
come variable (some researchers study how various experiences affect the
way people feel about themselves), and a mediating variable (self-esteem
needs are presumed to motivate a wide variety of psychological processes).
The widespread appeal of self-esteem attests to its importance, but this popu-
larity has had an undesirable consequence. Self-esteem has become a protean
conceptso capable of changing form that its value is in risk of being under-
mined.
Out purpose in this article is to explore the nature and functions of self-
esteem. We pay particular attention to the relation between self-esteem and the
way people evaluate themselves in specific domains. To set the stage for this
research, we begin by distinguishing three ways in which the term self-
esteem is used.
COGNITION AND EMOTION, 2001, 15 (5), 615631
Correspondenc e should be addressed to Jonathon D. Brown, Department of Psychology, Box
351525, University of Washington, Seattle, WA 98195-1525, USA; e-mail: jdb@u.washington.edu
This research was supported by a Presidential Young Investigator Award from the National
Science Foundation (SBR-8958211) and by a grant from the Horizons Foundation of Seattle.
# 2001 Psychology Press Ltd
http://www.tandf.co.uk/journals/pp/02699931.html DOI:10.1080/02699930143000004
Three meanings of self-esteem
Global self-esteem. Self-esteem is most commonly used to refer to the way
people characteristically feel about themselves. Many psychologists call this
form of self-esteem, global self-esteem or trait self-esteem, as it is relatively
enduring, both across time and situations. In the remainder of this paper, we will
use the term self-esteem (without any qualifiers) when referring to this
variable.
Attempts to define self-esteem have ranged from an emphasis on primitive
libidinal impulses (Kernberg, 1975), to the perception that one is a valuable
member of a meaningful universe (Solomon, Greenberg, & Pyszczynski, 1991).
We take a decidedly less exotic approach and define self-esteem in terms of
feelings of affection for oneself (Brown, 1993, 1998; Brown & Dutton, 1995).
High self-esteem is characterised by a general fondness or love for oneself; low
self-esteem is characterised by mildly positive or ambivalent feelings toward
oneself. In extreme cases, low self-esteem people hate themselves, but this kind
of self-loathing occurs in clinical populations, not in normal populations
(Baumeister, Tice, & Hutton, 1989).
Self-evaluations . The term self-esteem has also been used to refer to
the way people evaluate their various abilities and attributes. For example,
many scales designed to assess self-esteem include subscales for measuring
academic self-esteem, social self-esteem, or athletic self-esteem (Harter, 1986;
Marsh, 1990; Shavelson, Hubner, & Stanton, 1976). The terms self-
confidence and self-efficacy have also been used to refer to these beliefs,
and many people equate self-confidence with self-esteem. We prefer to call
these beliefs self-evaluations or self-appraisals, as they refer to the way
people evaluate or appraise their specific abilities and personality charac-
teristics.
Feelings of self-worth. Finally, self-esteem is also used to refer to rather
momentary emotional states, particularly those that arise from a positive or
negative outcome. This is what people mean when they speak of experiences
that bolster their self-esteem or threaten their self-esteem. For example, a person
might say her self-esteem was sky-high after getting a big promotion, or a
person might say his self-esteem plummeted after a divorce. Following William
James (1890), we refer to these emotions as self-feelings or feelings of self-
worth. Feeling proud or pleased with ourselves (on the positive side), or
humiliated and ashamed of ourselves (on the negative side) are examples of
what we mean by feelings of self-worth.
Many researchers use the term state self-esteem when referring to feelings of
self-worth (e.g., Butler, Hokanson, & Flynn, 1994; Leary, Tambor, Terdal, &
Downs, 1995). This term implies that the essential difference between global
616 BROWN, DUTTON, COOK
self-esteem and feelings of self-worth is that global self-esteem is more
enduring. We disagree with this approach. In our mind, global self-esteem and
feelings of self-worth are qualitatively different phenomena. To illustrate our
thinking here, consider that most parents swell with pride when their children do
something exemplary. But accomplishments of this sort do not change how
much love parents feel for their children. The pride comes and goes, often in
response to a particular event or achievement, but the love remains and is
independent of whether something has or has not been achieved. This is how we
think of the relation between global self-esteem and feelings of self-worth.
Feelings of self-worth rise and fall in response to particular outcomes, but global
self-esteem (or self-love) is enduring.
Relation among the three constructs
The theoretical distinction we have drawn regarding these uses of the term
self-esteem raises the question of how they are related. Although several
possibilities spring to mind, the research presented in this paper is guided by
an affective model of self-esteem functioning (Brown, 1993, 1998; Brown &
Dutton, 1995; Brown & Marshall, 2001; Dutton & Brown, 1997). In brief, the
affective model assumes that: (a) self-esteem develops early in life in
response to relational and temperamental factors; and (b) once formed,
endows high self-esteem people with the ability to promote, protect, and
restore high feelings of self-worth. This ability is particularly apparent when
people confront negative outcomes, such as failure in the achievement domain
or interpersonal rejection.
An investigation by Brown and Dutton (1995) provides some support for
these assertions. In this investigation, high self-esteem participants and low self-
esteem participants were randomly assigned to receive either success or failure
feedback on an alleged test of their intellectual ability. Afterward, all partici-
pants completed an eight-item emotion scale. Four of the items assessed general
feelings of happiness and sadness (happy, sad, glad, unhappy) and four of the
items assessed feelings of self-worth (proud, pleased with myself, humiliated,
ashamed). General feelings of happiness and sadness did not differ among the
two self-esteem groups, but high self-esteem participants reported higher feel-
ings of self-worth following failure than did low self-esteem participants. Along
with other research (Brown & Marshall, 2001; Dutton & Brown, 1997), these
results suggest that high self-esteem functions to promote and restore feelings of
self-worth.
High self-esteem people use a variety of strategies to protect and rebuild
feelings of self-worth (Brown, 1993, 1998). For example, rather than attributing
failure to low ability, they are more apt to say they failed because of lack of
effort or an ineffective strategy (Dutton & Brown, 1997). They also engage in
selective social comparison processes, comparing themselves with others when
FROM THE TOP DOWN 617
they are assured that the comparison will be favourable (Wood, Giordano-
Beech, & Ducharme, 1999). Following in this tradition, the present research
examines whether high self-esteem people also maintain high feelings of self-
worth by adjusting their self-evaluations in ways that lead them to believe they
possess socially desirable qualities (Baumeister, 1982; Brown & Smart, 1991;
Dodgson & Wood, 1998). In more formal terms, we hypothesise that: (a) self-
esteem influences self-evaluations; and (b) that self-evaluations influence feel-
ings of self-worth. These hypotheses join the three uses of the term self-
esteem we outlined earlier.
Overview of the present research
We conducted three investigations to test our hypotheses. In these studies, we
experimentally varied the purported favourability of a trait, and examined
whether self-esteem influences peoples willingness to endorse that trait as
self-descriptive. If, as we have claimed, self-esteem influences self-
evaluations, we ought to find that high self-esteem people are more apt to say
they possess a trait when it is described as desirable than when it is described
as undesirable.
Before presenting this research, we think its important to discuss the sig-
nificance of the experimental manipulation of trait desirability. Previous
research has established that self-esteem and self-evaluations are correlated
(e.g., Marsh, 1986, 1993, Pelham, 1995). This correlation could arise because:
(a) self-evaluations affect self-esteem; (b) self-esteem influences self-evalua-
tions; or (c) some unknown, third variable, influences self-evaluations and self-
esteem, without there being any causal link between them in either direction. By
experimentally manipulating the desirability of the trait, we are able to test the
viability of the second possibility. If self-esteem interacts with the purported
desirability of a trait to influence trait endorsement ratings, we would be war-
ranted in concluding that self-esteem (or some correlated third variable)
influences self-evaluations.
EXPERIMENT 1:
If its an important trait to have, ``I have it
Our first experiment examined the links between self-esteem, attribute impor-
tance, and self-evaluations. The participants were told that the experiment
concerned the measurement of a cognitive ability called integrative orientation.
Some participants were told this ability was important; others were told it was
unimportant. Participants then rated their integrative orientation ability. We
anticipated that high self-esteem participants would be more inclined than low
self-esteem participants to rate themselves more highly on the ability when it
was described as being important.
618 BROWN, DUTTON, COOK
METHOD
Participants
The participants were 103 University of Washington undergraduates. They
participated in exchange for extra course credit and were drawn from the top and
bottom thirds on Rosenbergs (1965) self-esteem scale. This ten-item scale is a
popular and well-validated measure of global self-esteem (Baumeister et al.,
1989; Rosenberg, 1979). Fifty-four participants were classified as having low
self-esteem (M=16.69); 49 were classified as having high self-esteem
(M= 27.78).
Materials and procedure
The experiment used a 2 (Self-esteem) 62 (Importance) experimental design.
Participants were tested in small groups of two to four students, and were seated
at a separate computer in such a way that they could not see one anothers
computer screens. All instructions and experimental materials were presented on
the computer.
At the start of the experiment, participants learned that the experiment
involved a problem-solving ability called integrative orientation. We described
integrative orientation as an aspect of creativity, an ability to find creative and
unusual solutions to problems. Using random assignment to conditions,
approximately half the participants learned that integrative orientation was a
very important ability to possess and the other half learned that psychologists
were not sure whether the ability was important or not, although no use for the
ability had yet been found. After receiving this information, participants indi-
cated: (a) how important they thought it was to be high in integrative orientation
ability (1 =not at all; 7 = very); and (b) then rated their integrative orientation
ability (1 = very low; 7 =very high). After completing these items, the partici-
pants were informed that the experiment was over. They were then debriefed,
thanked, and excused.
RESULTS AND DISCUSSION
We performed an initial analysis to determine whether the experimental
manipulation of importance was effective. A 2 62 (Self-esteem6Importance)
analysis of variance (ANOVA) on the importance question (How important do
you think it is to be high in integrative orientation?) yielded a single main
effect of importance, F(1, 99) =13.47, p < .001. Participants in the high impor-
tance condition thought that integrative orientation was a more important ability
(M= 5.0) than did participants in the low importance condition (M= 4.16).
We anticipated that the importance manipulations would alter participants
perceptions of whether or not they possessed the ability, with high self-esteem
participants being most inclined to say that they had the ability when they
FROM THE TOP DOWN 619
thought it was an important ability to possess. This proved to be the case. An
ANOVA on participants ability ratings produced significant main effects of
self-esteem, F(1, 99) =16.39, p <.001, and importance, F(1, 99) = 11.11,
p <.005, as well as the predicted Self-esteem6Importance interaction,
F(1, 99) =4.91, p <.05.
The nature of the interaction can be seen in Figure 1. In accordance with
predictions, high self-esteem participants rated their integrative orientation
ability more highly in the high importance condition than in the low importance
condition, F(1, 99) =15.40, p <.001; low self-esteem participants tended to do
the same, but the effect was not significant (F<1). It was also the case that high
self-esteem participants assumed that their ability was higher than did low self-
esteem participants when the ability was regarded as important, F(1, 99) = 19.63,
p <.001, but not when it was regarded as unimportant, F(1, 99) =1.68, n.s. In
effect, high self-esteem participants, but not low self-esteem participants, said:
If its an important ability to have, I have it.
EXPERIMENT 2:
If I have it, its important to have
Experiment 1 showed that high self-esteem people are particularly apt to lay
claim to possessing an important attribute. Experiment 2 tests a parallel effect.
In Experiment 2, we led some participants to believe they possessed a (fictitious)
ability and then asked them to indicate how important it is to possess the ability.
Figure 1. Mean ability ratings as a function of self-esteem and importance. (Error bars present
standard error of the mean.)
620 BROWN, DUTTON, COOK
We anticipated that high self-esteem participants would be especially apt to say:
If I possess the ability, its an important ability to possess.
METHOD
Participants
The participants were 106 University of Washington undergraduates. They
participated in exchange for extra course credit and were drawn from the top and
bottom thirds on Rosenbergs (1965) self-esteem scale. Fifty-four participants
were classified as having low self-esteem (M= 17l.54); 52 were classified as
having high self-esteem (M=27.98). The data from three other participants were
discarded due to mechanical problems.
Materials and procedure
The experiment used a 2 (Self-esteem) 62 (Ability) experimental design. As
in Experiment 1, participants were tested in small groups of two to four
students. All instructions and experimental materials were presented on the
computer.
At the start of the experiment, participants learned that the experiment
involved a problem-solving ability called integrative orientation. Instead of
manipulating importance (as we had done in Experiment 1), we simply asked
participants at this point to indicate how important they thought it was to be high
in integrative orientation ability (1 = not at all, 7 = very).
The experimental task was then introduced. This task was the Remote
Associates Test (Mednick, 1962). With this task, participants were shown three
words (e.g., carswimmingcue) and asked to find a fourth word that relates to
the other three (pool). Working interactively with the computer, participants
completed three sample problems to ensure that they understood how the
problems were solved.
Participants were then given 5 minutes to solve 10 problems. By varying the
difficulty of the problems they received, some participants were able to solve
many problems (high ability condition) and some were able to solve only a few
problems (low ability condition). Difficulty level was determined on the basis of
prior testing with an independent sample and on published norms (Brown &
Dutton, 1995; MacFarlin & Blascovich, 1984).
After the allotted time for working on the test had expired, the computer
paused for a moment and informed participants how many problems they had
correctly solved. The participants then: (a) evaluated their performance; (b)
rated their ability level; and (c) again indicated how important they thought it
was for a person to be high in integrative orientation ability. Finally, the par-
ticipants were informed that the experiment was over. They were then debriefed,
thanked, and excused.
FROM THE TOP DOWN 621
RESULTS AND DISCUSSION
Manipulation checks
Preliminary analyses indicated that the experimental manipulation was effective.
In comparison with those who received difficult problems, participants who
received easy problems: (a) solved more problems (Ms =7.35 and 3.96,
respectively); (b) evaluated their performance more favourably (Ms = 6.60 and
3.69, respectively); and (c) rated themselves higher in integrative orientation
ability (Ms =6.18 and 4.06, respectively; all ps < .001). There were no
Ability 6Self-Esteem interactions (all Fs <1).
1
Main analyses
To determine whether the experimental manipulations altered participants
perceptions of the importance of their integrative orientation ability, we per-
formed a 2 62 62 (Self-esteem6Ability 6Time: Pre-test Post-test) ANOVA,
with pre-test and post-test ratings of importance as a repeated measure. Two
effects reached significance: A main effect of time, F(1, 102) = 62.71, p < .001,
and the predicted three-way interaction, F(1, 102) =4.77, p < .05. The easiest way
to understand this interaction is to examine the bottom row of Table 1. These
change scores indicate that high self-esteem participants came to believe the
ability was more important when they had been led to believe they possessed it
than when they believed they lacked it, t(102) =2.39, p <.05; low self-esteem
participants tended to do the opposite, but the effect did not approach significance
(t < 1). Further analyses showed that the two self-esteem groups did not differ
when they were led to believe they lacked the ability (t <1), but high self-esteem
1
The only effect of self-esteem was a main effect of self-esteem on ability ratings. Across
experimental conditions, high self-esteem participant s claimed to have higher ability than did low
self-esteem participants (Ms = 6.2 and 4.05, respectively), F(1, 102) = 9.34, p <.001.
TABLE 1
Mean pre-task and post-task importance ratings as a function of self-
esteem and the experimental manipulation of ability as high or low:
Experiment 2
Low self-esteem High self-esteem
High ability Low ability High ability Low ability
Pre-task rating 4.06 4.07 4.32 4.40
Post-task rating 4.74 4.96 5.58 4.92
Change 0.68 0.89 1.24 0.53
Note: Values can range from 1 to 7.
622 BROWN, DUTTON, COOK
participants tended to believe the ability was more important than low self-esteem
participants when they believed they possessed it, t(102) = 1.86, p <.07. In effect,
high self-esteem participants, but not low self-esteemparticipants, said, If I have
the ability, its an important ability to have.
2
EXPERIMENT 3:
If its desirable to possess, I possess it
To this point we have seen that high self-esteem people: (a) claim to possess an
important trait; and (b) inflate the importance of traits they possess. We believe
these judgements represent an affect-management strategy, whereby high self-
esteem people arrange their self-evaluations to produce, maintain, and restore
high feelings of self-worth. One way to test this hypothesis is to examine how
trait endorsement decisions are influenced by threats to feelings of self-worth. If,
as we have argued, high self-esteem people use self-evaluations to promote
feelings of self-worth, their tendency to do so should be particularly evident
following an experience that undermines their feelings of self-worth, such as
interpersonal rejection or failure at an achievement-related task.
We examined this hypothesis in Experiment 3 by modifying a procedure
developed by Kunda and Sanitioso (1989). We presented participants with four
traits of ambiguous desirability (e.g., cautious, methodical). Half the participants
were told that the traits were desirable; the other half were told that the traits
were undesirable. Independently, half the participants had just succeeded on the
experimental task used in Experiment 2; the other had just failed. Finally, the
participants indicated to what extent they thought the traits described them.
Assuming that self-enhancement needs are stronger following a threat to self-
worth (Brown, Collins, & Schmidt, 1988; Brown & Gallagher, 1992; Brown &
Smart, 1991), and that self-evaluations reflect motivated needs to promote
feelings of self-worth, we expected that high self-esteem participants would be
most inclined to offer self-aggrandising self-evaluations after failure.
METHOD
Participants
The participants were 64 University of Washington undergraduates, participat-
ing for extra course credit. Using the same selection criteria as in the previous
two experiments, 36 of the participants had low self-esteem (M= 18.03) and 28
had high self-esteem (M= 27.54).
2
Another way to analyse these data is to use pretask importance scores as a covariate in a 2 62
(Self-esteem6Ability) analysis of covariance. This analysis produced a significant Self-esteem6
Ability interaction, F(1, 101) =4.75, p <.05, and the adjusted scores revealed a pattern of results
comparabl e to those reported in Table 1.
FROM THE TOP DOWN 623
Materials and procedures
The experimental procedure was similar to that used in Experiment 2 with one
exception. After learning how they had performed on the experimental task, the
participants were shown a list of four traits (cautious, methodical, single-
minded, strong-willed). Preceding the list was a brief paragraph describing the
traits as desirable or undesirable. The specific wording read:
The following are desirable (undesirable) qualities for a person to possess. Indi-
cate to what extent you think these desirable (undesirable) attributes describe
you.
The participants then rated the self-descriptiveness of each trait using 9-point
scales (1 =not at all, 9 = extremely). When they had finished making their rat-
ings, they signaled the experimenter to indicate that they were through. They
were then debriefed, thanked, and excused.
RESULTS
As in Experiment 2, participants who received easy problems: (a) solved more
problems; (b) evaluated their performance more favourably; and (c) rated
themselves higher in integrative orientation ability than did those who received
difficult problems (all ps <.001).
To test our main hypothesis, scores on the four traits were averaged to create
a self-evaluation index and were then analysed by means of a 2 62 62 (Self-
esteem6Test Performance 6Trait Desirability) ANOVA. The ANOVA
revealed main effects of Test Performance, F(1, 56) =5.56, p <.05, a Self-
esteem6Test Performance interaction, F(1, 56) =4.20, p .05, and a higher-
order Self-esteem6Test Performance 6Trait Desirability interaction, F(1, 56)
=4.25, p <.05.
Table 2 presents the means relevant to interpreting the higher-order effect. As
can be seen, high self-esteem participants, but not low self-esteem participants,
TABLE 2
Self-evaluations as a function of self-esteem, test performance, and
trait desirability: Experiment 3
Low self-esteem High self-esteem
Success Failure Success Failure
Desirable 6.14 5.81 6.70 6.38
Undesirable 5.85 6.00 6.69 4.44
Difference 0.29 70.19 0.01 1.94
Note: Values can range from 1 to 9.
624 BROWN, DUTTON, COOK
responded to failure by denying that undesirable traits described them. Simple
effects analyses revealed a Test Performance 6Trait Desirability interaction for
high self-esteem participants, F(1, 56) =5.46, p < .05, but not for low self-esteem
participants (F<1). Additional analyses revealed a Self-esteem6Trait Desir-
ability interaction following failure, F(1, 56) =6.63, p <.05, but not following
success (F <1). Finally, the mean for high self-esteem participants in the failure/
undesirable trait condition (M= 4.44) was significantly different than all of the
other group means (all ps < .05).
RESULTS AND DISCUSSION
The findings from Experiment 3 provide evidence that high self-esteem
people arrange their self-perceptions to promote positive feelings of self-
worth. Interestingly, they did so not by inflating the self-descriptiveness of
desirable traits, but by minimising the self-descriptiveness of undesirable
traits. This finding suggests that self-protection may be a more important con-
sideration to high self-esteem people than is self-promotion. Alternatively, the
particular traits we chose may be responsible for the effect. Despite our
attempts to manipulate the desirability of these traits, it is probably the case
that few people truly aspire to be methodical and single-minded. High self-
esteem people may simply have found it easier to restore feelings of self-
worth by denying they possess these negative traits than by claiming to
possess these positive traits.
GENERAL DISCUSSION
Self-esteem has been used to describe: (a) general feelings of affection for
oneself; (b) evaluations of self in specific domains; and (c) momentary feelings
of self-worth, such as pride and shame. In this paper, we investigated the
relations among these three constructs. Our research was guided by the notion
that high self-esteem people are more adept than low self-esteem people at
building and preserving high feelings of self-worth, and that they do so, in part,
by claiming to possess socially desirable traits and/or by denying that they
possess socially undesirable traits.
The findings from three investigations supported these hypotheses. In
Experiment 1 we found that high self-esteem participants claimed to possess an
ability when it was important to possess, and in Experiment 2 we found that high
self-esteem participants claimed that an ability they possessed was an important
one to have. Neither of these tendencies occurred among low self-esteem par-
ticipants. Finally, in Experiment 3 we found that high self-esteem participants
responded to failure by denying that (allegedly) undesirable traits described
them. Collectively, these findings are consistent with the claim that high self-
esteem people use their self-evaluations to promote and restore high feelings of
self-worth.
FROM THE TOP DOWN 625
Potential limitations
We see several important implications of our findings. Before discussing them,
we consider some potential limitations. One issue is whether our high self-
esteem participants truly believed their self-aggrandising self-assessments.
Some would argue that these judgements were offered primarily for public
consumption, shaped more by concerns with managing a public image of
competency than a desire to promote private feelings of self-worth (Baumeister
et al., 1989).
Several factors militate against this interpretation. First, all judgements were
made under anonymous conditions; at no time were participants asked to
identify themselves. Furthermore, the experimenter was not present when par-
ticipants made their judgements and participants made their responses working
interactively with a computer. These factors should have served to reduce the
public nature of the experimental setting in participants minds. Finally, and
perhaps most importantly, self-enhancing tendencies like those revealed in this
research have been shown to be at least as apparent under private conditions as
under public conditions (Brown, 1990; Brown & Gallagher, 1992). For these
reasons, we do not believe our findings can be understood with reference only to
impression management concerns.
Our use of extreme groups also merits comment. Instead of using only par-
ticipants with high self-esteem and low self-esteem, we might have included the
entire range of self-esteem scores and used a regression-based analytic approach,
rather than the ANOVA-based approach we utilised. We cannot say for sure how
this change would have affected our findings, but it is interesting to note that
prior research using continuous variables has produced a pattern of findings
compatible with the extreme-group, ANOVA approach used here (Brown &
Marshall, 2001; Dutton & Brown, 1997). Given this comparability, we see no
reason to assume on a priori grounds that our results would have been different
had an alternative analytic strategy been used.
It is also important to emphasise that our findings are correlational in nature.
Our experimental manipulations allow us to conclude that people classified as
high in self-esteem adjust their self-evaluations to match a traits purported
desirability, but we cannot rule out the possibility that the effects we have
attributed to self-esteem are due to some correlated third variable, such as
anxiety, depression, or negative affectivity (Watson & Clark, 1984). This
limitation is an inherent aspect of personality research.
Some researchers believe this problem can be overcome by experimentally
manipulating self-esteem (e.g., Arndt & Greenberg, 1999; Heatherton & Polivy,
1991; Leary et al., 1995). This is typically accomplished by giving people
positive or negative self-relevant feedback (e.g., telling people they are high or
low in some ability). This research strategy ignores the distinction we have made
between global self-esteem and feelings of self-worth. In our judgement,
626 BROWN, DUTTON, COOK
providing people with false personality feedback or varying their performance
on an achievement test influences their momentary feelings of self-worth but not
their self-esteem level. Because such manipulations do not provide a suitable
analogue for the experience of having high self-esteem or low self-esteem, they
do not overcome the limits of the correlational approach we have adopted.
Perhaps the best one can do is to measure variables that are known to correlate
with self-esteem, and control these variables in all statistical analyses (see
Brown & Marshall, 2001, for an illustration of this approach).
Implications
Despite these limitations, our findings have some important implications for
understanding the nature of self-esteem. A good deal of previous research has
shown a strong correlation between self-esteem and self-evaluations (Harter,
1986; Marsh, 1986, 1990, 1993, 1995; Pelham, 1995; Pelham & Swann, 1989).
This correlation has generally been taken to mean that self-evaluations deter-
mine self-esteem. Pelham and Swann (1989), for example, refer to self-eva-
luations as the building blocks of self-esteem (p. 673), arguing that people
move from specific knowledge of their abilities and accomplishments to global
[self-esteem] (p. 672).
The data we reported in this paper provide an alternative perspective on this
issue. Rather than viewing the relation between self-evaluations and self-esteem
as a bottom-up process, in which specific self-views determine global self-
esteem, our findings suggest that the relation may well be a top-down process, in
which self-esteem influences self-evaluations (Brown, 1993). From this per-
spective, global self-esteem guides the way people evaluate their specific
qualities. People who are fond of themselves in a general way (i.e., those with
high self-esteem), imbue themselves with many positive qualities. They like the
way they look; they enjoy their sense of humour; they appreciate their talents.
The causal process is a top-down one, from global feelings of affection to beliefs
that one possesses many socially valued attributes.
Although our data establish the viability of the top-down approach, they do
not rule out the possibility that bottom-up processes occur as well. Nevertheless,
we think there are good reasons to question the adequacy of the bottom-up
approach. For one thing, the notion that self-evaluations determine self-esteem
begs the question of what determines self-evaluations in the first place. Con-
sider, for example, peoples ideas about their attractiveness. At all ages, and for
both males and females, perceived attractiveness is closely related to self-esteem
(Harter, 1993; Pliner, Chaiken, & Flett, 1990). People who like the way they
look, like themselves (and people who like themselves, like the way they look).
The bottom-up approach assumes that the causal arrow goes from perceived
attractiveness to self-esteem: People somehow come to regard themselves as
attractive or unattractive and this decision affects their level of self-esteem.
FROM THE TOP DOWN 627
What this approach leaves unanswered is the question of why some people
regard themselves as attractive to begin with.
One possibility is that people correctly perceive how attractive they really
are, but this is not the case. A comprehensive review of the literature found that
the correlation between self-perceptions of attractiveness and attractiveness as
rated by others was .24 (Feingold, 1992). This value suggests that peoples
perceptions of their attractiveness are only weakly related to how attractive they
really are (as indexed by consensual judgements). Importantly, this value is not
low because observers disagree about who is attractive and who is not. Inter-
rater reliability (consensus among observers) in these studies is uniformly high,
typically exceeding .60. This means that people are in strong agreement about
the attractiveness of others but these consensual judgements do not coincide
with peoples perceptions of their own attractiveness.
This is by no means an isolated example. Although peoples self-perceptions
are rather accurate in domains of low importance (e.g., punctuality, con-
scientiousness), they are not very accurate in domains of high importance (e.g.,
intelligence, likeability). In fact, the more desirable the trait, the less accurate are
peoples self-evaluations (John & Robins, 1993; Park & Judd, 1989). In general,
then, the correspondence between what people actually are like and what they
think they are like in highly evaluative domains is modest at best.
These low associations would seem to pose problems for the bottom-up
approach. They tell us that peoples assessments of their specific qualities are
not unbiased, literal representations of what they are really like. Instead, they
appear to be schema-driven constructions and interpretations. The top-down
approach we favour can accommodate these findings. It states that peoples
perceptions of their specific qualities are heavily influenced by their overall
level of self-esteem. High self-esteem gives rise to the perception that one has
many positive qualities and is good at many things.
The malleability of self-evaluations creates another problem for the bottom-
up model. The bottom-up model assumes that self-esteem is built upon self-
evaluations, yet our findings show that high self-esteem people readily adjust
their self-evaluations to promote feelings of self-worth. The willingness with
which high self-esteem people modify their self-evaluations to promote feelings
of self-worth suggests that these evaluations might not provide the bedrock upon
which self-esteem rests.
A related issue concerns the shifting correlates of self-esteem at different
stages of life. During adolescence, self-esteem is closely related to beliefs about
ones popularity; in adulthood, self-esteem is linked to beliefs about ones
character and productivity. Using their self-evaluations to promote feelings of
self-worth, high self-esteem people assume that they possess those attributes that
are valued by their particular reference group at the time. The correlates of self-
esteem will therefore shift as people age, but these changes will represent shifts
in the manifestations of self-esteem rather than in its basis.
628 BROWN, DUTTON, COOK
Finally, our findings speak to the role self-evaluations play in self-esteem
functioning. Because it assumes that self-esteem depends on the way people
evaluate themselves in particular domains, the bottom-up approach holds that
self-esteem differences in behaviour are reducible to the way people view their
specific qualities. If we could equate the way people evaluate their specific
attributes, we would find no effect of global self-esteem.
Two investigations by Dutton and Brown (1997) recently examined this issue
and found mixed support for this position. Although self-evaluations (not self-
esteem) uniquely predicted peoples cognitive reactions to success and failure,
self-esteem (not self-evaluations) uniquely predicted peoples emotional reac-
tions to success and failure. In conjunction with the present findings, these
results support the claim that global self-esteem serves to regulate peoples
affective reactions to negative events. In our judgement, this is the primary
function of self-esteem: It allows people to fail without feeling bad about
themselves.
Manuscript received 9 March 2000
Revised manuscript received 15 November 2000
REFERENCES
Arndt, J., & Greenberg, J. (1999). The effects of a self-esteem boost and mortality salience on
responses to boost relevant and irrelevant worldview threats. Personality and Social Psychology
Bulletin, 25, 13311341.
Baumeister, R.F. (1982). Self-esteem, self-presentation, and future interaction: A dilemma of
reputation. Journal of Personality, 50, 2945.
Baumeister, R.F., Tice, D.M., & Hutton, D.G. (1989). Self-presentational motivations and person-
ality differences in self-esteem. Journal of Personality, 57, 547579.
Brown, J.D. (1990). Evaluating ones abilities: Shortcuts and stumbling blocks on the road to self-
knowledge. Journal of Experimental Social Psychology, 26, 149167.
Brown, J.D. (1993). Self-esteem and self-evaluation: Feeling is believing. In J. Suls (Ed.), Psy-
chological perspectives on the self (Vol. 4, pp. 2758). Hillsdale, NJ: Erlbaum.
Brown, J.D. (1998). The self. New York: McGraw-Hill.
Brown, J.D., Collins, R.L., & Schmidt, G.W. (1988). Self-esteem and direct versus indirect forms of
self-enhancement. Journal of Personality and Social Psychology, 55, 445453.
Brown, J.D., & Dutton, K.A. (1995). The thrill of victory, the complexit y of defeat: Self-esteem and
peoples emotional reactions to success and failure. Journal of Personality and Social Psy-
chology, 68, 712722.
Brown, J.D., & Gallagher, F.M. (1992). Coming to terms with failure: Private self-enhancement and
public self-effacement . Journal of Experimental Social Psychology, 28, 322.
Brown, J.D., & Marshall, M.A. (2001). Self-esteem and emotion: Some thoughts about feelings.
Personality and Social Psychology Bulletin, 5, 575584.
Brown, J.D., & Smart, S.A. (1991). The self and social conduct: Linking self-representations to
prosocial behavior. Journal of Personality and Social Psychology, 60, 368375.
Butler, A.C., Hokanson, J.E., & Flynn, H.A. (1994). A comparison of self-esteem liability and low
trait self-esteem as vulnerability factors for depression. Journal of Personality and Social Psy-
chology, 66, 166177.
FROM THE TOP DOWN 629
Dodgson, P.G., & Wood, J.V. (1998). Self-esteem and the cognitive accessibility of strengths and
weaknesses after failure. Journal of Personality and Social Psychology, 75, 178197.
Dutton, K.A., & Brown, J.D. (1997). Global self-esteem and specific self-views as determinants of
peoples reactions to success and failure. Journal of Personality and Social Psychology, 73, 139
148.
Feingold, A. (1992). Good-looking peopl e are not what we think. Psychological Bulletin, 111, 304
341.
Harter, S. (1986). Processes underlying the construction, maintenance, and enhancement of the self-
concept in children. In J. Suls & A.G. Greenwal d (Eds.), Psychological perspectives on the self
(Vol. 3, pp. 137181). Hillsdale, NJ: Erlbaum.
Harter, S. (1993). Causes and consequence s of low self-esteem in children and adolescents. In R.F.
Baumeister (Ed.), Self-esteem: The puzzle of low self-regard (pp. 87116). New York: Plenum.
Heatherton, T.F., & Polivy, J. (1991). Development and validation of a scale for measuring state self-
esteem. Journal of Personality and Social Psychology, 60, 895910.
James, W. (1890). The principles of psychology (Vol. 1). New York: Holt.
John, O.P., & Robins, R.W. (1993). Determinant s of interjudge agreement on personalit y traits: The
big five domains, observability, evaluativeness, and the unique perspective of the self. Journal of
Personality, 61, 521551.
Kernberg, O. (1975). Borderline conditions and pathological narcissism. New York: Jason
Alexander.
Kunda, Z., & Sanitioso, R. (1989). Motivated changes in the self-concept . Journal of Experimental
Social Psychology, 25, 272285.
Leary, M.R., Tambor, E.S., Terdal, S.K., & Downs, D.L. (1995). Self-esteem as an interpersonal
social monitor: The sociometer hypothesis. Journal of Personality and Social Psychology, 68,
518530.
Marsh, H.W. (1986). Global self-esteem: Its relation to specific facets of self-concept and their
importance. Journal of Personality and Social Psychology, 51, 12241236.
Marsh, H.W. (1990). A multidimensional , hierarchical model of self-concept : Theoretical and
empirical justification. Educational Psychology Review, 2, 77172.
Marsh, H.W. (1993). Relations between global and specific domains of self: The importance of
individual importance, certainty, and ideals. Journal of Personality and Social Psychology, 65,
975992.
Marsh, H.W. (1995). A Jamesian model of self-investment and self-esteem: Comment on Pelham
(1995). Journal of Personality and Social Psychology, 69, 11511160.
McFarlin, D.B., & Blascovich, J. (1984). On the Remote Associates Test (RAT) as an alternative to
illusory performance feedback: A methodological note. Basic and Applied Social Psychology, 5,
223229.
Mednick, S.A. (1962). The associative basis of the creative process. Psychological Review, 26, 220
232.
Park, B., & Judd, C.M. (1989). Agreement on initial impressions: Differences due to perceivers, trait
dimensions, and target behaviors. Journal of Personality and Social Psychology, 56, 493505.
Pelham, B.W. (1995). Self-investment and self-esteem: Evidence for a Jamesian model of self-worth.
Journal of Personality and Social Psychology, 69, 11411150.
Pelham, B.W., & Swann, W.B., Jr. (1989). From self-conceptions to self-worth: On the sources and
structure of global self-esteem. Journal of Personality and Social Psychology, 57, 672680.
Pliner, P., Chaiken, S., & Flett, G.L. (1990). Gender differences in concern with body weight and
physical appearance over the life span. Personality and Social Psychology Bulletin, 16, 263273.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University
Press.
Rosenberg, M. (1979). Conceiving the self. New York: Basic Books.
630 BROWN, DUTTON, COOK
Shavelson, R.J., Hubner, J.J., & Stanton, G.C. (1976). Self-concept: Validation of construct inter-
pretations. Review of Educational Research, 46, 407441.
Solomon, S., Greenberg, J., & Pyszczynski , T. (1991). A terror management theory of social
behavior: The psychological function of self-esteem and cultural worldviews. In M.P. Zanna
(Ed.), Advances in experimental social psychology (Vol. 24, pp. 93159). San Diego, CA:
Academic Press.
Watson, D., & Clark, L.A. (1984). Negative affectivity: The disposition to experience aversive
emotional states. Psychological Bulletin, 96, 465490.
Wood, J.V., Giordano-Beech, M., & Ducharme, M.J. (1999). Compensating for failure through social
comparison. Personality and Social Psychology Bulletin, 25, 13701386.
FROM THE TOP DOWN 631